Scaffold-free human vascular calcification model using a bio-three-dimensional printer

Morbidity and mortality rates associated with atherosclerosis-related diseases are increasing. Therefore, developing new research models is important in furthering our understanding of atherosclerosis and investigate novel treatments. Here, we designed novel vascular-like tubular tissues from multicellular spheroids composed of human aortic smooth muscle cells, endothelial cells, and fibroblasts using a bio-3D printer. We also evaluated their potential as a research model for Mönckeberg’s medial calcific sclerosis. The tubular tissues were sufficiently strong to be handled 1 week after printing and could still be cultured for 3 weeks. Histological assessment showed that calcified areas appeared in the tubular tissues within 1 week after culture in a medium containing inorganic phosphate (Pi) or calcium chloride as the calcification-stimulating factors. Calcium deposition was confirmed using micro-computed tomography imaging. Real-time quantitative reverse transcription polymerase chain reaction analysis revealed that the expression of osteogenic transcription factors increased in calcified tubular tissues. Furthermore, the administration of Pi and rosuvastatin enhanced tissue calcification. The bio-3D printed vascular-like tubular structures, which are composed of human-derived cells, can serve as a novel research model for Mönckeberg’s medial calcific sclerosis.


Introduction
From the latter half of the 20th century to the present, the incidence of diseases involving arterial calcification, such as ischemic heart disease and cerebral infarction, has increased worldwide, with a mortality rate of >15% [1]. In addition, due to the increasing elderly population, the morbidity of diseases involving arteriosclerosis is expected to continue increasing [2]. Therefore, developing novel research models that contribute to our understanding and investigation of new treatments for arteriosclerosis is important.
Arteriosclerosis causes thickening and stiffening of the arteries, as well as vascular remodeling [3], which includes atherosclerosis and nonatheromatous arteriosclerosis. In atherosclerosis, soft atheromas, which contains lipids, inflammatory cells, vascular smooth muscle cells (SMCs), thrombi, and calcium deposits with connective tissue, appears in the artery [4]. Advanced atherosclerosis causes coronary artery diseases, such as myocardial infarction, angina angiitis, or cerebral infarction [5]. In contrast, non-atheromatous arteriosclerosis includes arteriosclerosis and Mönckeberg's medial calcific atherosclerosis, which are characterized by intimal thickening and elastocalcinosis [6]. Furthermore, atherosclerosis is common in hypertension and diabetes mellitus and causes artery constriction, resulting in organ ischemia [7]. Mönckeberg's medial calcific atherosclerosis is associated with age-related degeneration of the vascular tunica media and calcified lesions in arteries, which are associated with diabetes mellitus and chronic renal disease, causing elevated pulse pressure and cardiac hypertrophy [7]. Atherosclerosis develops and progresses with aging and common diseases such as hypertension, diabetes mellitus, and dyslipidemia that affect a large number of patients. Once atherosclerosis has developed, effective treatment is limited and the risk of developing high mortality cardiovascular disease increases.
Therefore, to study the underlying mechanisms of atherosclerosis and to develop treatments, the following research models have been applied; in vitro models using cultured cells [8,9]; ex vivo models using animal blood vessels [10]; and in vivo models using genetically modified animals or disease model animals with feeding, substance application, or kidney removal [11]. However, the existing atherosclerosis study models have several issues. In vitro models frequently use single cultured cells, such as vascular SMCs, pericytes, or endothelial cells, which do not interact with other cells or the extracellular matrix. In many 3D models, cells are seeded and cultured on biomaterials for scaffolds comprising hydrogels or other materials [12,13]. These scaffolds may weaken cell-to-cell adhesion and inhibit tissue maturation. However, the aortic rings from rats and mice are commonly used for ex vivo studies, the vascular adventitia and media are in direct contact with calcification-inducing factors, and the luminal side of the flow is lost. Disease model animals are limited as they may have various comorbidities, such as hypertension and renal failure, making it difficult to study only atherosclerosis-related factors [14]. Therefore, we aimed to develop a new Mönckeberg's medial calcific atherosclerosis research model, considering the limitations of conventional experimental models of atherosclerosis.
We fabricated various scaffold-free structures on a bio-3D printer using a method developed in a previous study [15][16][17][18][19][20]. Our group also conducted clinical studies to implant a shunt graft [21,22] in patients undergoing hemodialysis therapy in whom an arteriovenous shunt could not be used because of shunt occlusion, stenosis, or aneurysm.
Here, we used this method to create tubular structures that mimic blood vessels using endothelial cells, vascular SMCs, and fibroblasts, which are components of arteries. This study aimed to investigate whether vascular-like structures could serve as a novel study model for Mönckeberg's medial calcific atherosclerosis.

Cells and cell culture
Human aortic SMCs (AOSMCs, CC-2571), human aortic endothelial cells (HAECs, CC-2535), and human normal dermal fibroblasts (NHDFs, CC-2509) were purchased from Lonza, Inc. (Walkersville, MD, USA). All cells were cultured following the manufacturer's instructions. AOSMC, HAEC, and NHDF were cultured in SmGM-2 (CC-3182, Lonza), EGM-2 (CC-3162, Lonza), and FGM-2 (CC-3132, Lonza) with growth supplement (Lonza) media, respectively. AOSMCs were used between passages 8 and 10, and HAECs and NHDFs were used up to passage 7 in this study. The cell instruction manual states that all cells can be used up to passage 15 and extended culture of each cell could be performed without issue.

Generation of scaffold-free tubular-like tissue using a bio-3D printer
Tubular-like tissues were constructed using a bio-3D printer (Regenova. Cyfuse Biomedical K.K., Tokyo, Japan) [20]. First, multicellular spheroids were collected with a suction nozzle (O.D. of 0.45 mm and I.D. of 0.23 mm), next, placed with a robotically controlled arm onto the needles of a pin holder, and formed into a computer-designed 3D tubular shape. The pinholder have stainless needles, with a diameter and distance of 0.17 mm and 0.4 mm, respectively, between the needles. After printing, the tubular structure was incubated in a culture medium (SmGM-2, EGM-2, and FGM-2 at a ratio of 1:1:1) with the pin holder and perfused using a roller pump at 2 ml min −1 for 7 d. After the spheroids were fused, the tubular structure was removed from the pin holder (figures 1(a) and (b)).

Circulation chamber and cultures of circulation
The circulation chamber was fabricated using a polycarbonate box with inflow and outflow ducts, and a 16-gauge plastic catheter (Terumo, Tokyo, Japan) was connected to the ducts (figure 1(c)). Polydimethylsiloxane (PDMS) tubes (O.D. of 2 mm; Kyushu Semiconductor KAW Co., Ltd, Fukuoka, Japan) were connected to plastic catheters, and the bio-3D printed tubular-like structure removed from the pinholder was transferred to the PDMS tube. At this point, the culture medium was changed to Dulbecco's Modified Eagle Medium (DMEM) (low glucose, Fujifilm, Tokyo, Japan) supplemented with 10% fetal bovine serum (FBS, Sigma-Aldrich, MO, USA) and 1% penicillin-streptomycin (Fujifilm), perfused by a roller pump at 3 ml min −1 . Finally, tubular-like structures were maintained at 37 • C in a humidified incubator with 5% CO 2 .

Supplements for induction of calcification
A calcification induction medium was developed by including inorganic phosphate (Pi) and calcium chloride (Ca) in DMEM supplemented with 10% FBS and 1% penicillin-streptomycin. Pi was prepared by mixing Na 2 HPO 4 and NaH 2 PO 4 solutions (Nacalai Tesque, Inc.. Kyoto, Japan), whose pH was adjusted to pH 7.4 and added to the serum-supplemented DMEM to a final concentration of 2.9 mmol l −1 . Finally, 0.1 mol l −1 Ca solution (Nacalai Tesque) was added to the supplemented DMEM to a final concentration of 2.8 mmol l −1 .

Drug administration
Rosuvastatin calcium (Fujifilm) was dissolved in dimethyl sulfoxide (Nacalai Tesque) and added to the supplemented DMEM at 0.1 µmol l −1 . An aliquot of dimethyl sulfoxide was added to the nonrosuvastatin-administered group.

Histochemistry, immunohistochemistry, immunofluorescence and quantification of calcification area
Briefly, tissues were fixed in 4% paraformaldehyde (Fujifilm, Tokyo, Japan), and the water in the tissue was replaced with a sucrose solution, embedded in an optimal cutting temperature compound (Sakura, Tokyo, Japan), and frozen in isopentane cooled with liquid nitrogen. Next, the frozen tissues were sliced to a thickness of 12 µm thickness using a cryostat (Leica Biosystems).
Hematoxylin and eosin, alizarin red, and von Kossa staining, and Elastica van Gieson staining were performed using conventional methods. Ca deposition was evaluated using light microscopy (BZ-X700, Keyence, Osaka, Japan) for samples without nuclear counterstaining, and the calcified area was measured using analysis software (BZ-H3A, Keyence).
In immunofluorescence, the sliced sections were blocked and permeabilized with blocking solution overnight at 4 • C. Primary antibodies, which include anti-human von Willebrand Factor (vWF) antibody (F3520, Sigma, 1:100), anti-α-smooth muscle actin antibody (A2547, Sigma, 1:200) in blocking solution for overnight at 4 • C. The sections were washed three times with phosphate-buffered saline, and incubated with appropriate Alexa Fluor 488-, 594-conjugated secondary antibodies (A11005, A11008, ThermoFisher Scientific, USA) diluted 1:500 in blocking solution for 90 min at room temperature, washed three times, and mounted on slides with ProLong Diamond antifade mountant with DAPI (P36962, ThermoFisher Scientific, USA). Images were captured using a microscope (BZ-X700; Keyence) and using analysis software (BZ-H3A, Keyence). Apoptosis was detected through the terminal deoxynucleotidyl transferase-mediated dUTP nick end labeling (TUNEL) assay using the in situ Apoptosis Detection Kit (MK500, Takara) and stained according to the manufacturer's instructions.

Quantitative real-time reverse transcript polymerase chain reaction (qRT-PCR)
Gene expression in tubular-like tissues was evaluated using qRT-PCR. Total RNA was isolated using an RNeasy Plus Universal Mini Kit (73404; QIAGEN, Hilden, Germany) following the manufacturer's protocol. cDNA was also synthesized using QuantiTect Reverse Transcription (205311, QIAGEN), per the manufacturer's protocol. qRT-PCR was subsequently performed using TaqMan Fast Advanced Master Mix (Applied Biosystems) and target TaqMan probes (Applied Biosystems, MA, USA) with QuantStudio 3 (Applied Biosystems). Gene expression was calculated relative to the endogenous control (beta actin) mRNA in each sample to obtain the relative quantification value. The comparative CT method (2-∆∆CT) was used to calculate the average fold difference between the samples. The probes used for the qRT-PCR are summarized in table 1.

Statistical analysis
All experiments were performed in a minimum of three replicates. The significance of more than three groups was analyzed using a one-way analysis of variance (ANOVA) with Tukey's post hoc test using R (Ver 3.3.3.). The data are expressed as mean ± standard deviation (SD). Statistical significance was set at P < 0.05.

Calcification occurred in bio-3D printed tubular tissues after being cultured in a calcification induction medium
Multicellular spheroids were prepared from AOSMCs, HAECs, and NHDFs. The average diameter of the multicellular spheroids used for 3D printing the tubular structures was 580 µm. Tubularlike structures were fabricated using a bio-3D printer ( figure 2(a)), and he spheroids on 96-well plates were picked up by suction with a nozzle and skewered onto a pin holder one at a time with a robotic arm [16].
The multicellular spheroids with a cell ratio of 70% of AOSMCs, 20% HAECs, and 10% NHDFs were used for the following reasons: AOSMCs were most abundant as they were thought to play an important role in vascular calcification; HAECs were second in abundance after AOSMCs, however spheroids with more than 30% HAECs were too soft and therefore difficult to print, thus 20% HAECs were used. NHDFs were required to contain at least 10% to maintain the strength of the structure.
The tubular structures were removed from the pinholder 7 d after printing and were subsequently mounted on PDMS tubes in the circulation chamber to be cultured with a culture medium containing Pi (2.9 mmol l −1 ) and Ca (2.8 mmol l −1 ) ( figure 2(b)). PDMS is reported to be highly oxygen permeable, maintaining a good oxygen supply to the constructs and making it suitable for culture [24,25]. After perfusion, micro-CT imaging was performed at 1, 2, and 3 weeks ( figure 3(a)). In the control group, the tissue of the tubular structure could not be identified using micro-CT. However, micro-CT showed a highdensity area in the tissues in the Pi-or Pi-and Caadded groups 1 week after culture in the calcification induction medium. The high-density area occurred along the entire structure and was considered tissue calcification. Moreover, the calcified area expanded with a longer culture period, and the mean CT number increased.
Frozen sections were prepared from these tubular structures, and alizarin red and von Kossa staining were performed to quantitatively evaluate the calcified areas (figures 3(b) and (c). Similar to the micro-CT results, tissue calcification was not observed in the control group; however, highly calcified lesions appeared in the interior and surface of the tubular structures in the calcification induction medium groups. Furthermore, the calcified areas, which increased with longer cultivation periods, were measured using von Kossa staining images in the calcification induction medium groups ( figure 3(d)). Von Kossa staining showed no appearance of tissue calcification in the control group even on day 21; however, in the Pi-added group, 8.1, 27.9, and 27.2% of calcified area were observed on day 7, day 14, and day 21, respectively. In Pi and Ca-added group, 12, 23.3, and 23.4% calcification was observed on day 7, day 14, and day 21, respectively.
AOSMCs and HAECs were maintained in the tissue without cell loss even after 3 weeks of long-term culture ( figure 4(a)). However, AOSMCs tended to accumulate outside the structure, while HAECs accumulated mostly on the luminal surface and inside the structure ( figure 4(b)). Elastica van Gieson staining  revealed that there were no elastic fibers in the tubular structure; however, collagen fibers appeared in tissue sections (supplementary figure 1). As culture continued on day 21 after printing (calcification day 14) and on day 28 after printing (calcification day 21), red collagen fibers gradually appeared on the inside and on the surface of the structures (supplementary figure  1). Immunofluorescence staining showed that some

Gene expression of osteogenic markers increased in calcified tubular structures
In biological calcified arteriosclerosis, expression of osteogenic genes is upregulated, and the SMC phenotype changes to osteoblast-like cells [26]. Aging, inflammation, oxidative stress, hyperphosphatemia, and oxidized low-density lipoprotein (LDL)-stimulated arteries can induce runt-related transcription factor 2 (RUNX2) and Msh homeobox 2 (MSX2) osteogenic transition of vascular SMC [27,28]. These transcription factors induce the expression of the bone morphogenic protein (BMP) and that of other transcription factors, resulting in vascular calcification. In addition, the expression of vascular SMC markers (α-actin and SM-22α) decreases during this process [29].
Here, total RNA was prepared from tubular structures cultured in the calcification induction or control medium, cDNA was synthesized, and qRT-PCR was performed ( figure 4(c)). Compared with the control group, the MSX2 expression was significantly upregulated in the calcification induction medium groups from the early stages of culture, whereas those of RUNX2 and BMP2 increased after the second week of culture. The RT-qPCR results showed that MSX2 was upregulated 1.9-fold in Pi-the treated group and 1.7fold in the Pi + Ca-treated group on day 7 compared to the control. RUNX2 was upregulated by 1.5-fold in the Pi-treated group on day 7, 1.2-fold on day 7, and 1.8-fold on day 14 in the Pi + Ca-treated group, compared to that in the control. The BMP2 expression increased by 2.3-fold in the Pi-treated group on day 14 and 1.3-fold on day 21 in the Pi + Catreated group, compared to the control. These results showed that osteoblast-like transformation occurred in vascular SMCs in the calcified tubular structures and indicated active calcification in tubular tissues. In the Pi-and Pi + Ca-treated groups, RUNX2 expression decreased to the same level as that of the control group in the third week after the start of culture in the calcification induction medium. However, micro-CT images, alizarin red staining, and von Kossa staining showed that tissue calcification was most enhanced in the third week of culture. These data suggest that RUNX2 initiates the osteoblast-like transformation of vascular SMCs in the early stages of vascular calcification in vivo.

High concentration of rosuvastatin stimulated to increase tissue calcification
Next, we evaluated whether the drug administration altered the calcification status of the tubularlike structures. Statins are effective for the secondary prevention of arteriosclerosis; several clinical studies have reported that strong statins, such as rosuvastatin, pitavastatin, and atorvastatin, effectively reduce plaques in the vascular intima [30]. In vitro studies have also demonstrated that apoptosis and calcifium deposition in vascular SMCs increases with increasing concentration of Pi in the medium [31], whereas low concentrations of statins inhibit Pi-induced vascular SMC apoptosis and calcification [32]. Here, we administered rosuvastatin, which is widely used in clinical practice, simultaneously with the calcification induction medium at the same concentration at the end of the study ( figure 5(a)).
After culturing in the calcification-inducing medium, the tubular-like structures were more intensely calcified in the Pi and rosuvastatin-treated groups than in the Pi-treated group, as shown by micro-CT imaging, because the CT numbers of Pi and rosuvastatin tissues were elevated ( figure 5(b)). Von Kossa staining showed that the calcified area of the Pi and rosuvastatin-treated groups were increased compared with that of the Pi-treated group; however, the difference was not statistically significant ( figure 5(c)). In the Pi and rosuvastatin-treated group, 13.2, 26.6, and 27.9% of areas appeared to be calcified on day 7, day 14, and day 21, respectively. In addition, Pi was administered to induce tissue calcification, followed by the administration of rosuvastatin, there was a trend towards fewer calcified areas that were positive for alizarin red staining (supplementary figure 3).
qRT-PCR results showed that MSX2, BMP2, and RUNX2 expression levels were elevated in the Pi and rosuvastatin-treated groups to the same extent as those in the Pi-treated group ( figure 5(d)). In the Pi and rosuvastatin-treated group, MSX2 expression increased by 2.1-fold, BMP2 expression increased by 1.2-fold, and RUNX2 expression increased by 1.5fold on day 7, compared to those in the control. The TUNEL assay showed strong TUNEL-positive staining in the calcified areas of Pi and rosuvastatin-treated tissues, and the degree of apoptosis in both groups was comparable (figure 5(e)).

Discussion
There are many diseases with an underlying cause of arteriosclerosis, such as stroke, cardiovascular diseases, and chronic renal failure; therefore, establishing a novel study model is crucial in investigating arteriosclerosis pathophysiology and treatment. Here, vascular-like tubular structures were fabricated from cultured human-derived cells containing vascular SMCs, which play an important role in vascular calcification [33], using a bio-3D printer. We examined whether this structure could be used as a new research model for Mönckeberg's medial calcific sclerosis. However, no reports of arteriosclerosis research models created similarly exist. These vascular-like structures reached easy handling strength 7 d after bio-3D printing and could be cultured for as long as 3 weeks. Since no exogenous components were present, the cell-to-cell adhesion of this structure was sufficiently firm, and it did not affect the penetration of the medium into the tissues nor the drug administration [34]. Although imperfectly formed, many endothelial cells were found on the lumen side of the structure and vascular SMC clustered on the outer lumen side, which is similar to in vivo blood vessels. Calcification-inducing medium supplemented with Pi or Ca induced tissue calcification in tubular-like structures in as little as 1 week, and RUNX2, MSX2, and BMP2 expression was increased, similar to those in in vivo arteries [35].
We also administered rosuvastatin to the culture medium to investigate whether rosuvastatin could inhibit Pi-induced calcification. Micro-CT imaging and histological evaluation with alizarin red and von Kossa staining showed no reduction in the calcified areas in tubular-like tissues in the Pi and rosuvastatin-treated groups compared with the Pitreated group. The Pi and rosuvastatin-treated groups tended to increase in the calcified area, although this was not statistically significant. Moreover, qRT-PCR results showed a similar trend. MSX2, BMP2, and RUNX2 expression levels were slightly higher in the Pi and rosuvastatin-treated groups than those in the Pi-treated group. These results could be due to rosuvastatin-induced apoptosis and the increased tissue apoptosis, which enhanced tissue calcification. However, statins inhibit Pi-induced apoptosis at low concentrations but increase apoptosis at high concentrations [36]. Tissue apoptosis can cause vascular calcification [37]. In clinical studies on humans, repeated oral administration of 20 mg rosuvastatin resulted in a blood concentration of approximately 0.02 µmol l −1 [38]. In this study, rosuvastatin was administered at a concentration of 0.1 µmol l −1 in reference to previous in vitro studies, which is approximately five times higher than the human blood concentration of 20 mg rosuvastatin. We fabricated vascular-like structures from human cells, and a low concentration of rosuvastatin < 0.1 µmol l −1 would have been suitable. Therefore, we plan to conduct future studies using lower concentrations of rosuvastatin. Note that AOSMCs with a passage number of 8-10 were used in this study, because AOSMCs with passages less than 7 did not calcify even with Pi administration (supplementary figures 4(a) and (b)). One of the molecular mechanisms of vascular calcification is the decreased expression of nicotinamide adenine dinucleotide-dependent protein deacetylase sirtuin 1 (SIRT1) in aged cells [39]. Depletion of cellular SIRT1 can cause cellular calcification of AOSMCs, leading to the calcification of tissues.
The culture method used in this study was the perfusion culture of vascular-like structures on PDMS tubes. However, this method differs from in vivo methods since calcification-inducing factors are directly in contact with the outside of the structure, and there is no inner flow, which is a limitation of our study. In future studies, we plan to conduct experiments more similar to in vivo blood vessels, where both ends of the vascular-like structure are fixed and perfused through the lumen. Our laboratory confirmed that organ models fabricated from multicellular spheroids using a bio-3D printer had the appropriate medium flow, and each cell migrated to a location similar to that of the in vivo organ. Although the incubation of vascular-like structures with the inner flow may cause endothelial cells to accumulate on the inside of the structures, vascular SMCs and fibroblasts may accumulate on the outer side. Therefore, th is structure is assumed to be more similar to an in vivo vessel.
We used Pi and Ca, which are commonly employed during in vitro studies, as calcificationinducing factors. The physiological serum phosphate level is 3-4.5 mg dl −1 (0.97-1.46 mmol l −1 ) [40]; however, patients with chronic kidney disease may have developed hyperphosphatemia, i.e. serum phosphate level exceeding 9 mg dl −1 (2.92 mmol l −1 ) [41,42]. Thus, in the present study, we added 2.9 mmol l −1 of Pi, which was considered an appropriate phosphate concentration for the disease model. The physiological serum calcium level is 4.5-5.6 mg dl −1 (1.1-1.4 mmol l −1 ), and the condition in which the serum calcium levels exceed 10.5 mg dl −1 (2.6 mmol l −1 ) is defined as hypercalcemia [43]. In this study, we added 2.8 mmol l −1 of calcium, equivalent to mild hypercalcemia, in the Ca-added group; this is considered an appropriate calcium concentration for the disease model. Next, we examined other substances that induce calcification. However, several drugs inhibit vascular calcification in vitro and in vivo [44][45][46][47][48][49]. In this study, we performed rosuvastatin administration experiments but did not confirm the effect of inhibiting tissue calcification. It is assumed that if the calcification of tubular-like structures can be inhibited or improved through drug administration, it will pave the way for development of novel treatments for arteriosclerosis. For example, TGF-β may be effective in reducing tissue calcification and inhibiting the progression of atherosclerosis [50,51]. Therefore, we plan to examine the administration of various agents and the culture conditions.

Conclusion
The bio-3D printed vascular-like structures composed of human-derived cells were sufficiently strong to be handled in a short period of time, could be cultured for a long time, and showed active calcified areas in the tissue after a short period of calcification induction. Therefore, in future studies, we will develop this technology and fabricate more identical in vivo arteries by perfusing the inner tubular structures. Furthermore, the vascular-like structures used in this study represent a potential option for a new research model of Mönckeberg medial calcific sclerosis research model.

Data availability statement
All data that support the findings of this study are included within the article (and any supplementary files).